Before reading starts, please click "Follow", not only can discuss and share, but also bring a different sense of experience, thank you for your support!
Text|Broken Boat Han
Editor|Broken Boat Han
preface
Over the past few decades, the global trend in dairy farming has been to automate and robotize the milking process on commercial farms, with a common average of 3.5 lactation periods for cows.
Due to the rapid development of dairy farming, there is an increasing need for new technologies to achieve higher economic efficiency and increase production.
On the one hand, intensive production increases milk production in cows, and on the other hand, intensive production leads to rapid deterioration of cows, i.e. less frequent lactation.
The shortening of the productive life of dairy cows also depends on the premature culling of animals with high or low physical condition scores, and abnormal physical condition scores during lactation are mainly due to dietary deficiencies.
Another negative consequence is the culling of animals due to poor physical condition, as an increase in physical condition scores reduces fertility and thus prolongs the service period.
The Physical Condition Score (BCS) assessment is important in technical milk production, first of all, the BCS score is used to place animals in the productivity group and determine their status.
In Russian dairy farm conditions, veterinarians and animal husbandry technicians rotate animals once a month, provided that milk production technology is mature.
The Body Condition Score helps to make decisions individually based on the current physiological condition of each cow, not just based on accepted technical specifications.
So it leads to a natural decrease in the physical condition score, experts in a study showed that during calving, the physical condition score decreases, the body fat reserve increases milk production, and by 40-65 days after breastfeeding, as much milk as possible, the physical condition score should not decrease by more than 0.5.
During this period, the cows consume up to 12 kg of high-energy feed, insemination occurs when the animal is at the peak of production, and the amount of feed consumed is the highest and the fat score is within 3 points, and the animal's physical condition score must not deviate by more than ±0.25 points.
Devices and systems
The image of the cow's back was collected on the 3D commercial camera O3D3033DToF camera, the system is powered by 220V, the power supply is converted to 24V to ensure the normal operation of the 3D sensor, the sensor is mounted at a height of +2.2m above the ground.
Demonstrates how the distance of the scanned surface varies with the angle of inclination of the sensor.
The height and tilt angle of the 3D sensor are based on four parameters: cow height, cow length, minimum working distance between camera and object, and allowable error of the camera, and the height of the animal under study is between 1300 mm and 1500 mm.
The minimum working distance between the camera and the object under study is 300 mm, and the signal from the animal RFID tag that identifies the antenna triggers the generation of a three-dimensional image.
Taking into account the given 4 parameters, the tilt angle of the sensor is chosen by 5 degrees, as it can cover a sufficient area of the back of the animal being analyzed.
While keeping the pixel pitch at 0.006m, when the point of interest is away from the 3D camera, the distance between pixels is 0.006m, which is the set distance based on the least squares method when forming a cluster of points related to the area of interest.
Number of images and cow ratio scored for each physical condition.
The total dataset consisted of 182 images from 546 animals, with physical condition scores ranging from 1 to 5 and a step size of 0.25 points, according to earlier studies, this camera took three photos of each cow was enough, and then the images were combined and the system began to determine fat.
From the data collected, it can be seen that the main physical condition scores of 4.75, 4.25, 3.25, 3, 2.5, 2.25, 2, 1.5 The distribution of animals by physical condition score was carried out by the panel of experts, whose opinion was considered a benchmark.
A device for field data collection was developed.
Physical condition score assessment and outcome analysis
Using previously developed software to process the obtained 3D maps and determine physical condition scoring and criteria tools for primary data processing, and formatted Excel for processing research results.
Results are obtained automatically and the results of expert assessments are manually compared, and the assessment of the expert's physical condition score is a baseline value.
In terms of searching and identifying the main area of interest, the software developed is based on the application of the least squares method to find the area of interest.
Since the camera is mounted on top of the animal and presents the points on the cow's back to the data analysis, the points of most interest are near the outline and describe the points around it.
1 - filter area; 2—the silhouette of the cattle; 3—spine of unspecified animals; 4—the spine of the specified animal; 5—tail; 6—Buttocks.
In the case of the spine of a cow, it can be identified considering basic expressions, and the algorithm developed is based on ordinary least squares.
According to the relevant data, the results of the regression method show the ridge structure as a set of points on each vertical axis of the construction.
In order to determine the height of the cow, it is necessary to estimate the coordinates of each point along the ridgeline and find the extreme value, which is the withering that determines the height of the cow.
Two methods of BCS assessment, the upper half is used by a panel of experts for BCS assessment and the lower half of the form is used for automated assessment.
The panel of experts uses automated methods to evaluate the method of determining the area of interest for the cow's physical condition score.
The values are determined manually by analyzing the resulting animal field database, and the numerical characteristics are the average of each physical condition score, related to the black variegated and Holstein black variegated breeds raised in central Russia, for other breeds the numerical characteristics may differ.
In the hip area, the angle is systematically evaluated: two lines are drawn along the protrusion of the back, and then the angle is evaluated, with an angle of 136° indicating a physical condition score of 3 and an angle of 125°> indicating an obesity score of 1.75 to 1.
The desired area of interest.
As the last step before determining fat, the system checks all criteria and determines the physical condition score on a 25-point scale on a step size of 5.0 points.
Converted into three-dimensional images in black and white format, these pictures show the animal's physical condition score from 1 to 5 out of 5 and explain which area of the animal's back the algorithm manipulated.
Physical condition score of the animal and areas of interest 1. Low back/dorsal spinous process; 2. Transverse process in the hollow area of waist hunger; 3. Hip bones and needle bones; 4. Caudal head 5. Vulvar and area
outcome
In order to understand the overall effectiveness of the development system, it is necessary to analyze and evaluate the effectiveness of each area of interest, evaluate all generated field data using the developed methodology, compare the results with the evaluation of experts.
In order to understand the overall effectiveness of the developed system, it is necessary to analyze the evaluation effectiveness of each area of interest.
The efficiency of the system when detecting the area of interest of the animal under study.
The results of the correlation analysis showed that the proposed method was able to estimate the area of the tail floor with 100% accuracy, the accuracy rate of hunger cavities was 98.9%, the determination rate of vulvar and regions was 95.10%, and the evaluation accuracy of protruding vertebrae (i.e., spinous and transverse processes) was 52.20% and 51.10%.
The 50% accuracy is due to the fact that these regions are not identified or identified incorrectly in animals with a physical condition score range, 3.25 to 5, and experts estimate that the overall accuracy of this system is 93.4%, which is a positive result.
Distribution plot of 5-6 months old animal fat obtained in automated assessment.
In observation, most of the animals with a fat score of 3.75-5 were in the second farm, with an average annual milk production of 15 kg/milk per cow per day and a fat content of 5.1-6%.
This farm is in good financial condition, the main activity is to obtain milk from animals for cheese production, when analyzing the causes of overweight animals, it is found that the farm staff is disrupting the feeding diet, and the animals are getting more energy than they need; The animals are kept in loose housing.
Plot of fat distribution of 6-32 months old animals obtained during automated assessment of 5 animals.
In the second farm, there was a correlation between the live weight of 32 animals and the physical condition score, and the correlation determined by the Pearson method was R=0.849.
The first part of the BCS algorithm.
According to the results of the study, the algorithm for automatic assessment of the animal's physical condition (fat) is supplemented and modified, and then its physiological condition is staged, the algorithm is divided into two parts, the second part of the algorithm is the component of the first part, the algorithm is included in the software code of the development software.
The second part of the BCS algorithm.
Technical suitability
After confirming the cost-effectiveness of the developed BCS estimation technique, it is now possible to describe how to implement automated BCS assessment for milk production.
Automatic livestock monitoring systems, which operate in two ways, the first way is stationary, the second way is mobile, and the stationary method includes the fact that it is on the farm.
For example, where animals pass daily, the BCS evaluation system is installed behind the milking parlor in the "gallery".
The system consists of a three-dimensional camera and data collection, and processing unit and an identification antenna that reads the cow's ID number and the data is sent to the server.
Schematic of installing a mobile BCS evaluation system in a compartment housing a group of animals.
The mobile method means that the system is taken to a box where a group of animals is kept every month to scan the BCS score.
Based on the data obtained, the developed software draws a chart - BCS, a graph of the change in the score and compares it with the set required values of the current physiological state of each cow, the technique has several applications, the first is when there is an increase in the physical condition score of an animal.
The system records an increase in the BSC score and then queries the herd management software for the following data: current physiological state, which group the animal belongs to.
Current milk production, lactation day, insemination status and specific diet, for example, an animal on the 75th day of lactation, not conceived, BCS score of 3.75, milk production of 17 kg/day, fat rate of 3.7%.
It is then automatically decided that the diet should be adjusted by reducing the amount of energy the animal receives, without changing the animal's maintenance group, as the animal is at the peak of lactation and should meet its milking requirements.
When moving to the next group, a gradual decrease in milking should be observed accordingly, while the condition of the animal must be monitored so that at the end of lactation, the condition of the animal is corrected.
For example, an animal on day 190 of lactation and day 110 of conception has a BCS status score of 2.5, milk production of 14 kg/day, and fat content of 3.5%.
In this case, animals should be moved from group 3 to group 1 or group 2 to adjust the feeding level to ensure excess energy.
conclusion
By implementing the Automatic Dairy Herd Fat Assessment Technology (BCS), which is 25.5 steps on a 0-point scale, the effect of increasing dairy farm production has been achieved, and the developed technology has been tested on 3 pig farms with a total of 1,810 cows.
A contactless BCS assessment of the required dairy herd was carried out over the entire life cycle of the herd on the farm, and the overall accuracy of the system was estimated at 93.4%, and experts demonstrated the economic effectiveness of implementing the proposed system.
If you also like my article, you may wish to click "follow"! Thank you here!
END